Re: [PATCH v18 16/17] x86/resctrl: Sub NUMA Cluster detection and enable

[Reinette Chatre <reinette.chatre@intel.com>]

Hi Tony,

On 5/15/2024 3:23 PM, Tony Luck wrote:
> There isn't a simple hardware bit that indicates whether a CPU is
> running in Sub NUMA Cluster (SNC) mode. Infer the state by comparing

Sometimes it is "Sub-NUMA Cluster" and sometimes it is "Sub NUMA Cluster".

> number CPUs sharing the L3 cache with CPU0 to the number of CPUs in
> the same NUMA node as CPU0.
> 
> When SNC mode is detected, reconfigure the RMID counters by updating
> the MSR_RMID_SNC_CONFIG MSR on each socket as CPUs are seen.
> 
> Clearing bit zero of the MSR divides the RMIDs and renumbers the ones
> on the second SNC node to start from zero.
> 
> Signed-off-by: Tony Luck <tony.luck@intel.com>
> ---
>  arch/x86/include/asm/msr-index.h   |   1 +
>  arch/x86/kernel/cpu/resctrl/core.c | 104 +++++++++++++++++++++++++++++
>  2 files changed, 105 insertions(+)
> 
> diff --git a/arch/x86/include/asm/msr-index.h b/arch/x86/include/asm/msr-index.h
> index e022e6eb766c..3cb8dd6311c3 100644
> --- a/arch/x86/include/asm/msr-index.h
> +++ b/arch/x86/include/asm/msr-index.h
> @@ -1164,6 +1164,7 @@
>  #define MSR_IA32_QM_CTR			0xc8e
>  #define MSR_IA32_PQR_ASSOC		0xc8f
>  #define MSR_IA32_L3_CBM_BASE		0xc90
> +#define MSR_RMID_SNC_CONFIG		0xca0
>  #define MSR_IA32_L2_CBM_BASE		0xd10
>  #define MSR_IA32_MBA_THRTL_BASE		0xd50
>  
> diff --git a/arch/x86/kernel/cpu/resctrl/core.c b/arch/x86/kernel/cpu/resctrl/core.c
> index dd40c998df72..195f9e29c553 100644
> --- a/arch/x86/kernel/cpu/resctrl/core.c
> +++ b/arch/x86/kernel/cpu/resctrl/core.c
> @@ -21,6 +21,7 @@
>  #include <linux/err.h>
>  #include <linux/cacheinfo.h>
>  #include <linux/cpuhotplug.h>
> +#include <linux/mod_devicetable.h>
>  
>  #include <asm/cpu_device_id.h>
>  #include <asm/resctrl.h>
> @@ -753,11 +754,42 @@ static void clear_closid_rmid(int cpu)
>  	      RESCTRL_RESERVED_CLOSID);
>  }
>  
> +/*
> + * The power-on reset value of MSR_RMID_SNC_CONFIG is 0x1
> + * which indicates that RMIDs are configured in legacy mode.
> + * This mode is incompatible with Linux resctrl semantics
> + * as RMIDs are partitioned between SNC nodes, which requires
> + * a user to know which RMID is allocated to a task.
> + * Clearing bit 0 reconfigures the RMID counters for use
> + * in Sub NUMA Cluster mode. This mode is better for Linux.
> + * The RMID space is divided between all SNC nodes with the
> + * RMIDs renumbered to start from zero in each node when
> + * couning operations from tasks. Code to read the counters

couning -> counting

> + * must adjust RMID counter numbers based on SNC node. See
> + * __rmid_read() for code that does this.
> + */
> +static void snc_remap_rmids(int cpu)
> +{
> +	u64 val;
> +
> +	/* Only need to enable once per package. */
> +	if (cpumask_first(topology_core_cpumask(cpu)) != cpu)
> +		return;
> +
> +	rdmsrl(MSR_RMID_SNC_CONFIG, val);
> +	val &= ~BIT_ULL(0);
> +	wrmsrl(MSR_RMID_SNC_CONFIG, val);
> +}
> +
>  static int resctrl_arch_online_cpu(unsigned int cpu)
>  {
>  	struct rdt_resource *r;
>  
>  	mutex_lock(&domain_list_lock);
> +
> +	if (snc_nodes_per_l3_cache > 1)
> +		snc_remap_rmids(cpu);
> +
>  	for_each_capable_rdt_resource(r)
>  		domain_add_cpu(cpu, r);
>  	mutex_unlock(&domain_list_lock);
> @@ -997,11 +1029,83 @@ static __init bool get_rdt_resources(void)
>  	return (rdt_mon_capable || rdt_alloc_capable);
>  }
>  
> +/* CPU models that support MSR_RMID_SNC_CONFIG */
> +static const struct x86_cpu_id snc_cpu_ids[] __initconst = {
> +	X86_MATCH_VFM(INTEL_ICELAKE_X, 0),
> +	X86_MATCH_VFM(INTEL_SAPPHIRERAPIDS_X, 0),
> +	X86_MATCH_VFM(INTEL_EMERALDRAPIDS_X, 0),
> +	X86_MATCH_VFM(INTEL_GRANITERAPIDS_X, 0),
> +	X86_MATCH_VFM(INTEL_ATOM_CRESTMONT_X, 0),
> +	{}
> +};
> +
> +/*
> + * There isn't a simple hardware bit that indicates whether a CPU is running
> + * in Sub NUMA Cluster (SNC) mode. Infer the state by comparing the
> + * number CPUs sharing the L3 cache with CPU0 to the number of CPUs in
> + * the same NUMA node as CPU0.
> + * It is not possible to accurately determine SNC state if the system is
> + * booted with a maxcpus=N parameter. That distorts the ratio of SNC nodes
> + * to L3 caches. It will be OK if system is booted with hyperthreading
> + * disabled (since this doesn't affect the ratio).
> + */
> +static __init int snc_get_config(void)
> +{
> +	struct cpu_cacheinfo *ci = get_cpu_cacheinfo(0);
> +	const cpumask_t	*node0_cpumask;
> +	cpumask_t *l3_cpumask = NULL;
> +	int i, ret;
> +
> +	if (!x86_match_cpu(snc_cpu_ids))
> +		return 1;
> +
> +	cpus_read_lock();
> +	if (num_online_cpus() != num_present_cpus())
> +		pr_warn("Some CPUs offline, SNC detection may be incorrect\n");
> +	cpus_read_unlock();
> +
> +	for (i = 0; i < ci->num_leaves; i++) {
> +		if (ci->info_list[i].level == 3) {
> +			if (ci->info_list[i].attributes & CACHE_ID) {
> +				l3_cpumask = &ci->info_list[i].shared_cpu_map;
> +				break;
> +			}
> +		}
> +	}
> +	if (!l3_cpumask) {
> +		pr_info("can't get CPU0 L3 mask\n");

Sentence can start with upper case

> +		return 1;
> +	}
> +
> +	node0_cpumask = cpumask_of_node(cpu_to_node(0));
> +
> +	ret = bitmap_weight(cpumask_bits(l3_cpumask), nr_cpu_ids) /
> +	       bitmap_weight(cpumask_bits(node0_cpumask), nr_cpu_ids);
> +

Can cpumask_weight() be used?

> +	/* sanity check: Only valid results are 1, 2, 3, 4 */
> +	switch (ret) {
> +	case 1:
> +		break;
> +	case 2 ... 4:
> +		pr_info("Sub-NUMA cluster detected with %d nodes per L3 cache\n", ret);
> +		rdt_resources_all[RDT_RESOURCE_L3].r_resctrl.mon_scope = RESCTRL_NODE;
> +		break;
> +	default:
> +		pr_warn("Ignore improbable SNC node count %d\n", ret);
> +		ret = 1;
> +		break;
> +	}
> +
> +	return ret;
> +}
> +
>  static __init void rdt_init_res_defs_intel(void)
>  {
>  	struct rdt_hw_resource *hw_res;
>  	struct rdt_resource *r;
>  
> +	snc_nodes_per_l3_cache = snc_get_config();
> +
>  	for_each_rdt_resource(r) {
>  		hw_res = resctrl_to_arch_res(r);
>  

Reinette